Appeal 2007-0114
Application 10/990,960
We have reviewed the example at para. [004]. We are unable to see
how this example shows predicting price movement based on a majority of
output values being the same, let alone “substantially” the same. Para.
[0040] discusses output signals, not output values. As we understand it (see
para. [0039]), a neural network model has an output value (Y) which ranges
from about -0.2 to 1.2. The output value is transformed into a discrete
integer score, called an N-score, ranging from 1-20. A particular score
determines which output signal - a buy (B), sell (S), or neutral (N) output
signal - is generated for the respective model (see para. [0037]). Para [0040]
discusses a neural network comprising a number of models, each generating
a B, S, or N output signal. Predictions using the network are based on which
output signal, B, S, or N, garners the majority of models.
To argue, as Appellants have, that the specification at para. [0040]
exemplifies predicting price movement based on a majority of output values
rather than output signals, presumes that every model’s output value
undergoes the same transformation in generating a corresponding output
signal. That has not been established. There is insufficient information to
11
Page: Previous 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Next
Last modified: September 9, 2013